As p-channel and n-channel field effect transistors in complementary metal oxide semiconductor integrated circuits have been scaled down, reliability problems with these transistors have increased.
In n-channel field effect transistors (NFETs) as the channel lengths have decreased, a transistor degradation sometimes referred to as a hot-electron effect has become a greater problem. The hot-electron effect has been studied for quite some time now. Power supplies were accordingly reduced as the transistors were scaled down in order to minimize the hot-electron effect. Designers can design around this somewhat by increasing the channel length of a transistor or increase signal slew rates. For example the channel lengths of certain transistors that may experience greater stress, such as output drivers, may be increased over that of the minimum channel lengths, in order to decrease the hot electron effect.
In the p-channel field effect transistors (PFETs), a new transistor degradation has more recently been discovered. This phenomenon is sometimes referred to as “negative bias temperature instability” (NBTI) but may more commonly be referred to as PMOS BT (p-channel metal oxide semiconductor field effect transistor (“PMOS”) bias temperature) or PBT as it causes the turn on threshold (Vth) of the PFET to shift.
In order to ascertain whether integrated circuit designs and semiconductor process can overcome these transistor degrading effects, the integrated circuit was put through quality assurance and reliability testing.
Typical quality assurance and reliability testing for integrated circuits was to “burn-in” or run the functional device for extended periods of time in hot ovens at higher voltages and then test or characterize the functionality of the integrated circuit over the corners of the power supply, operating temperature, and clock speed to see if it would still function. However, “burn-in” is not actually how an integrated circuit is used in a system. The integrated circuit will experience other conditions such as power cycling, large temperature variations, and even physical vibration while in use.
In other cases, dedicated test integrated circuits are used to determine the quality assurance and the reliability of a given semiconductor process. These dedicated test integrated circuits that provide characterization information, have little to no functionality and are typically designed for experimental testing only. In order to obtain a measure of the quality assurance and the reliability, the dedicated test integrated circuit may be similarly “burned-in” and then tested or characterized over the corners of the power supply and operating temperature. In this case measurements are taken to try and determine the reliability and quality of a process and a design. But again, this is not actually how an integrated circuit is used in a system.